Subtitles for accessibility

The subtitles most of us use casually were developed, first and foremost, as an accessibility technology. Understanding that history changes how you look at them.

The deaf community and early captions

Captions for deaf audiences were pioneered long before they became common in mainstream distribution. In the United States, the first captioning experiments date to the 1950s — the Captioned Films for the Deaf program, funded by an act of Congress, brought subtitled educational films to deaf schools. These were hard-coded captions burned into the film. They were the beginning of what would eventually become a national captioning infrastructure.

In the 1970s, the Public Broadcasting Service (PBS) began experimenting with broadcast captions using a technique called "open captioning" — captions that were visible to all viewers, burned into the broadcast. This was followed by "closed captioning" — a separate signal that could be decoded by a set-top box, allowing captions to be turned on or off.

The closed caption decoder became commercially available in 1980. Television manufacturers were required to include caption decoders in all televisions sold in the US by 1993. This was a landmark accessibility mandate and a model for future regulation.

Captions vs subtitles

The terminology matters:

• Subtitles translate dialogue. They assume the viewer can hear the audio, just not understand the language.
• Captions or closed captions transcribe dialogue and describe non-dialogue audio for viewers who cannot hear. They include sound effects, speaker identification and music cues.

For most of broadcast history, "captions" meant accessibility and "subtitles" meant translation. The line has blurred in streaming, where "subtitles" often means either.

The three critical contributions of accessibility subtitles

Subtitles developed for accessibility contributed three things to the broader subtitle world:

1. Non-dialogue audio descriptions

SDH (Subtitles for the Deaf and Hard of Hearing) introduced the practice of describing sound effects in brackets: [door slams], [phone rings], [distant thunder]. This technique is now used in mainstream subtitle contexts when the audio matters to plot comprehension.

2. Speaker identification

SDH guidelines require marking which character is speaking, especially when the identity is ambiguous: - MARY: Hello, John.. This convention has become standard in professional subtitling even for general-audience files, because it improves comprehension in scenes with multiple speakers.

3. Reading-speed limits

SDH subtitles are the origin of the 12–17 CPS targets that now influence all subtitle style guides. Because deaf viewers have no audio cue for pacing — they must read the subtitle and process the dialogue entirely from the text — the reading speed must be more conservative than for hearing audiences.

Regulatory milestones

• 1990: Americans with Disabilities Act (ADA) in the US. Interpreted to require accessible public accommodations, including captions.
• 1993: FCC mandates built-in caption decoders in televisions.
• 1996: Telecommunications Act requires video programming distributors to provide closed captions.
• 1998: Section 508 mandates accessible technology for federal agencies.
• 2008: Twenty-First Century Communications and Video Accessibility Act (CVAA) extends captioning to online video that was originally broadcast on TV.
• 2018: WCAG 2.1 specifies captions as required for pre-recorded content.
• 2019: EU Web Accessibility Directive extends accessibility obligations to public sector websites and apps.
• 2025: European Accessibility Act (EAA) applies to many industries, including e-commerce and consumer services.

The trend is clear: captioning is moving from a public-sector requirement to a universal expectation.

Streaming era and the quality problem

Streaming platforms have invested significantly in captioning because regulatory and accessibility requirements apply there too. Netflix, Amazon and Disney+ each publish SDH style guides that are more rigorous than most broadcast standards.

But the streaming era also introduced a new problem: auto-generated captions. ASR-driven captions are fast and cheap. They are also often wrong. When a platform ships uncorrected ASR captions as if they were accessibility-quality captions, deaf viewers get a degraded experience. This has become a serious accessibility issue.

Professional captioners argue that auto-captioning is not accessibility. It is a first draft at best. Treating it as a final deliverable for deaf audiences is a form of second-class service.

SDH subtitle design

Creating good SDH subtitles is a craft. Some principles:

• Preserve meaning, not words. Compress dialogue to fit the reading-speed target. If a line cannot fit, use a shorter paraphrase rather than dropping the line.
• Label speakers when ambiguous. Not every line needs a label; only lines where the speaker is not obvious from context.
• Describe significant sound effects. Not every sound. Background hum does not need captioning. A gunshot, a phone ring or a doorbell does.
• Coordinate with on-screen text. If a sign says "CLOSED" on screen, the SDH caption should not duplicate it, but should indicate its presence: [sign: CLOSED].
• Respect timing. SDH cues need slightly more lead-out time because the additional text (brackets, speaker labels) increases reading load.

The accessibility mindset

The accessibility mindset is simple: assume the viewer cannot hear. What information do they need to understand the story? Dialogue is only part of it. Tone of voice, off-screen sound, music cues, the identities of speakers — all of it matters.

Subtitles built with that mindset are not just compliant — they are better. They serve deaf viewers, non-native speakers, viewers in noisy environments, viewers in silent environments and viewers who simply pay better attention when reading.

Tools that support accessibility subtitles

• Line Length Checker — ensure lines fit within SDH guidelines.
• Duration Report — verify cue durations support slower SDH reading speeds.
• SRT to VTT — prepare captions for HTML5 delivery, which is what most accessibility audits check.
• SRT Validator — validate before publishing.